EE325 Probability and Random Processes Sets and set operations; Probability space, Conditional probability and Bayes

theorem,Combinatorial probability and sampling models, Discrete random variables, probability mass function, probability distribution function, example random variables and distributions, Continuous random variables, probability density function, probability distribution function, example distributions, Joint distributions, functions of one and two random variables, moments of random variables, Conditional distribution, densities and moments, Characteristic functions of a random variable, ar!ov, Chebyshev and Chernoff bounds; "andom se#uences and modes of convergence $everywhere, almost everywhere, probability, distribution and mean s#uare%, &imit theorems, Strong and wea! laws of large numbers, central limit theorem'"andom process' Stationary processes' ean and covariance functions'(rgodicity' )ransmission of random process through &)*' Power spectral density' Text/References +'Star! and J' ,oods,--Probability and "andom Processes with .pplications to Signal Processing,// )hird (dition, Pearson (ducation' $*ndian (dition is available%'

.' Papoulis and S' 0nni!rishnan Pillai, --Probability, "andom 1ariables and Stochastic Processes,// 2ourth (dition, c3raw +ill' $*ndian (dition is available%' 4' &' Chung, *ntroduction to Probability )heory with Stochastic Processes, Springer *nternational Student (dition' P' 3' +oel,S' C'Port and C'J' Stone *ntroduction to Probability, 0BS Publishers,l P' 3' +oel, S' C' Port and C' J' Stone, *ntroduction to Stochastic Processes,0BS Publishers S' "oss, *ntroduction to Stochastic odels, +arcourt .sia, .cademic Press'

EE601 Statistical Signal Analysis "eview of probability theory and random variables5 )ransformation $function% of random variables, Conditional expectation;Se#uences of random variables5 convergence of se#uences of random variables';Stochastic processes5 wide sense stationary processes, orthogonal increment processes, ,iener process, and the Poisson process, 4& expansion'; (rgodicity, ean s#uare continuity, mean s#uare derivative and mean s#uare integral of stochastic processes'; Stochastic systems5 response of linear dynamic systems $e'g' state space or ." . systems% to stochastic inputs, &yapunov e#uations, correlational function, power spectral density function, introduction to linear least s#uare estimation, ,iener filtering and 4alman filtering' Text/References

.' Papoulis, Probability, "andom 1ariables and stochastic processes, 6nd (d',

c3raw +ill, 789:'

.' &arson and B';' Schubert, Stochastic Processes, 1ol'* and **, +olden<Day, 78=8' ,' 3ardener, Stochastic Processes, c3raw +ill, 789>'

EE60 Ada!ti"e Signal Processing "eview of linear and non<linear estimation theory' Signal modelling' ;ptimal filtering';.daptive filtering as an extension of the optimal least mean s#uare error case .daptive algorithms5 adaptive e#uali?ation and echo cancellation; adaptive lattice filters' .pplication to radar, sonar, geophysics and hydrology, economic processes, communications $spread spectrum techni#ues%' Text/References

S' +ay!in, .daptive filter theory, Prentice +all, 789>' B' ,idrow and S'D' Stearns, .daptive signal processing, Prentice +all, 789@'

EE610 #mage Processing *mage representation < 3ray scale and colour *mages, image sampling and #uanti?ation';)wo dimensional orthogonal transforms < D2), 22), ,+), +aar transform, 4&), DC)';*mage enhancement < filters in spatial and fre#uency domains, histogram<based processing, homomorphic filtering';(dge detection < non parametric and model based approaches, &;3 filters, localisation problem';*mage "estoration < PS2, circulant and bloc! < circulant matrices, deconvolution, restoration using inverse filtering, ,iener filtering and maximum entropy<based methods'; athematical morphology < binary morphology, dilation, erosion, opening and closing, duality relations, gray scale morphology, applications such as hit<and<miss transform, thinning and shape decomposition';Computer tomography < parallel beam proAection, "adon transform, and its inverse, Bac!<proAection operator, 2ourier<slice theorem, CBP and 2BP methods, ."), 2an beam proAection';*mage communication < JP(3, P(3s and +'6>x standards, pac!et video, error concealment';*mage texture analysis < co<occurence matrix, measures of textures, statistical models for textures' isc' topics such as < +ough )ransform, boundary detection, chain coding, and segmentation, thresholding methods' Text/References

.' 4' Jain, 2undamentals of digital image processing, Prentice +all of , 7898' "' ' +aralic!, and &'3' Shapiro, Computer and "obot 1ision, 1ol<7, .ddison ,esley, , 7886' "' Jain, "' 4asturi and B'3' Schunc!, achine 1ision, c3raw<+ill *nternational (dition, 788B'

,' 4' Pratt, Digital image processing, Prentice +all, 7898' .' "osenfold and .' C' 4a!, Digital image processing, 1ols' 7 and 6, Prentice +all, 789>' +' C' .ndrew and B' "' +unt, Digital image restoration, Prentice +all, 78==

EE621 $ar%o" &'ains and ()e)ing Systems Prere#uisite5 Bac!ground in Probability and Stochastic Processes and an interest in System odeling'; ar!ov Chains and regenerative processes have been extensively used in modeling a wide variety of systems and phenomena' &i!ewise, many systems can be modeled as #ueueing systems with some aspect of the #ueue governed by a random process' ;bvious examples of such systems occur in telecommunication systems, manufacturing systems and computer systems' )his course is aimed at teaching system modeling using ar!ov chains with special emphasis on developing #ueueing models' )he course contents are as follows';*ntroduction5 "eview of basic probability, properties of nonnegative random variables, laws of large numbers and the Central &imit )heorem';"enewal Processes5 Basic definitions, recurrence times, rewards and renewal reward theorem, point processes, Poisson process, ,alds e#uation, Blac!well/s theorem';Discrete time ar!ov chains5 definitions and properties, matrix representation, Perron<2robenius theory';Continuous time ar!ov chains5 basic definitions, C<matrix, birth<death processes, #uasi birth death processes';(mbedded ar!ov processes, semi ar!ov processes, reversible ar!ov chains';"andom wal!s';2undamental #ueueing results5 &ittle/s theorem, invariance of the mean delay, Conservation law'; ar!ovian #ueues5 Jac!son and BC P networ!s, numerical .lgorithms'; D3D7 E 3D D7 #ueues and 3D3D7 #ueues;.dvanced #ueueing models5 priority, vacation and retrials in #ueues' Text/References

Stochastic P'Bremaud,

odeling and the )heory Cueues, Prentice +all, Cliffs, 7898' ar!ov Chains, Springer<1erlag, 7888' atrices and ar!ov Chains, Springer Series in Statistics, Springer, ,

('Seneta, Fon Fegative 7897'

"'3allager, Discrete Stochastic Processes, 4luwer .cademic Press, 788>' &'4leinroc!, Cueueing SystemsG vols * and **, John ,iley and Sons 78=>'

EE635 A!!lied *inear Algebra in Electrical Engineering 1ector spaces, linear dependence, basis; "epresentation of linear transformations with respect to a basis';*nner product spaces, +ilbert spaces, linear functions; "ies? representation theorem and adAoints';;rthogonal proAections, products of proAections, orthogonal direct sums; 0nitary and orthogonal transformations, complete orthonormal sets and Parseval/s identity; Closed subspaces and the proAection theorem for +ilbert spaces';Polynomials5 )he algebra of polynomials, matrix polynomials, annihilating polynomials and invariant subspaces, forms';.pplications5 Complementary orthogonal spaces in networ!s, properties of graphs and their relation to vector space properties of their matrix representations; Solution of state e#uations in linear system theory; "elation between the rational and Jordan forms';Fumerical linear algebra5 Direct and iterative methods of solutions of linear e#uations; atrices, norms, complete metric spaces and complete normal linear spaces $Banach spaces%; &east s#uares problems $constrained and unconstrained%; (igenvalue problem' Text/References

4' +offman and "' 4un?e, &inear .lgebra, Prentice<+all $%, $789>%' 3'+' 3olub and C'2' 1an &oan, atrix Computations, .cademic, 789:'

3' Bachman and &' Farici, 2unctional .nalysis, .cademic Press, 78>>' ('4reys?ig, ntroductory functional analysis with applications John ,iley, 78=9'

EE636 $atrix &om!)tations Basic iterative methods for solutions of linear systems and their rates of convergence' 3enerali?ed conAugate gradient, 4rylov space and &anc?os methods' *terative methods for symmetric, non<symmetric and generali?ed eigenvalue problems' Singular value decompositions' 2ast computations for structured matrices' Polynomial matrix computations' Perturbation bounds for eigenvalues' Text/References

;we .xelsson, *terative solution methods' Cambridge'788@' 3' eurant5 Computer solution of large linear systems' Forth +olland, 7888' atrix computations' John +op!ins Press 788>' atrix perturbation theory' .cademic press,788H'

3olub and C' 1an &oan5 3' ,' Stewart and J' Sun5

EE63 Estimation and #dentification

*ntroduction to linear least s#uare estimation 5 a geometric approach',iener filter, &evinson filter, updating C" filter and the 4alman filter'2ilter implementation structures 5 lattice, ladder and the systolic C"'Stochastic reali?ation theory $modelling given the covariance%' odelling given the raw data' Spectral estimation';"ecursive least s#uares identification algorithms5 &evinson<type, 4alman<type and the C"<type' Text/References

0'B' Desai, &ecture notes on estimation reali?ation and identification, 0npublished, 789>' B'D'; .nderson and J'B' oore, ;ptimal 2iltering, Prentice<+all, 78=8'

)' 4ailath, &ecture notes on ,iener and 4alman 2iltering, Springer 1erlag, 789H' &' &Aung, System *dentificaiton )heory for the user, Prentice<+all, 789=' P'S' aybec!, Stochastic odels, (stimation and Control, 1ols' 7<:, .cademic Press, 789H<7896'

EE65+ A ,irst &o)rse in -!timi.ation otivation' mathematical review , matrix factori?ations, sets and se#uences, convex sets and functions, linear programming and simplex method, ,eierstrass/ theorem, 4arush 4uhn )uc!er optimality conditions, algorithms, convergence, unconstrained optimi?ation, &ine search methods, method of multidimensional search, steepest descent methods, Fewton/s method, modifications to Fewton/s method , trust region methods, conAugate gradient methods, #uasi<Fewton/s methods' constrained optimi?ation, penalty and barrier function methods, augmented &agrangian methods, polynomial time algorithm for linear programming, successive linear programming, successive #uadratic programming' Text/References

"' 2letcher Practical ;ptimi?ation $6nd (dition% John ,iley E Sons, Few Ior!, 789=' 'S'Ba?araa , +'D'Sherali and C'Shetty , Fonlinear Programming, )heory and .lgorithms, John ,iley and Sons, Few Ior!, 788:'

EE6/ 0a"elets *ntroduction to time fre#uency analysis; the how, what and why about wavelets';Short<time 2ourier transform, ,igner<1ille transform';Continuous time wavelet transform, Discrete wavelet transform, tiling of the time<fre#uency plane and wavepac!et analysis';Construction of wavelets' ultiresolution analysis' *ntroduction to frames and biorthogonal wavelets'; ultirate signal processing and filter ban! theory';.pplication of wavelet theory to signal denoising, image and video compression, multi<tone digital communication, transient detection' Text/References

I')' Chan, ,avelet Basics, 4luwer Publishers, Boston, 788:' *' Daubechies, )en &ectures on ,avelets, Society for *ndustrial and .pplied Philadelphia, P., 7886' athematics,

C' 4' Chui, .n *ntroduction to ,avelets, .cademic Press *nc', Few Ior!, 7886' 3erald 4aiser, . 2riendly 3uide to ,avelets, Bir!hauser, Few Ior!, 788B' P' P' 1aidyanathan, ultirate Systems and 2ilter Ban!s, Prentice +all, Few Jersey, 788:'

.'F' .!ansu and "'.' +addad, ultiresolution signal Decomposition5 )ransforms, Subbands and ,avelets, .cademic Press, ;ranld, 2lorida, 7886' B'Boashash, )ime<2re#uency signal analysis, *n S'+ay!in, $editor%, .dvanced Spectral .nalysis, pages @79<<B7=' Prentice +all, Few Jersey, 7887'

EE/02 &om!)ter 1ision Pre<re#uisite5 ((<>7H';*maging model and geometry5 scene radiance and image irradiance, reflectance model of a surface, &ambertian and specular reflectance, photometric stereo';*ll<posedness of vision problems5 regulari?ation theory';Shape from shading, structured light and texture' ;ptical flow, structure from motion and recursive motion analysis' Stereo vision and correspondence problem';Depth analysis using real<aperture camera5 depth from defocused images'; "2 approach to early vision problems5$shape from shading, matching, stereo and motion%, *mage texture analysis';*ntroduction to image understanding';*ntegrated vision, sensor fusion' Text/References

B' 4' P' +orn, "obot 1ision, D'

*) Press, 789>'

arr, 1ision, 2reeman and Co', San 2rancisco, 7896'

S' Chaudhuri and .' F' "aAagopalan, Depth from Defocused *mages, Springer 1erlag, FI, 7888'Selected Papers'

EE/10 *arge S!arse $atrix &om!)tations Sparse atrices, applications in electrical engineering, data structures, linear system solvers, ordering5 ar!owit? criterion, minimum degree E minimum deficiency ordering for sparse symmetric positive definite matrices, enhancements to D., sparse indefinite linear system solvers, eigen value computation, sparse matrix optimi?ation, applications in power systems5 state estimation, load flow E optimal power flow'

EE//+ Ad"anced To!ics in Signal Processing *ntroduction < "elevant concepts in DSP, linear algebra, matrix analysis, and statistical signal analysis $as needed% Spectral estimation $temporal% < Fon<parametric spectrum estimation < Parametric methods for rational spectra $." ., ., ." processes% and line spectra, .rray signal processing $spatial% < .rrays and spatial filters, space<time processes < ,aveform estimation < broadband and narrowband beamformers < Subspace algorithms < 0S*C, (SP"*), "oot< 0S*C < .pplications < direction<of<arrival $D;.% estimation, signal separation, Selected

topics in sampling and reconstruction < Dithered sampling, use of a random dither < Sampling, #uanti?ation, and interpolation < Sampling of non<bandlimited signals < 2inite rate of innovation signals, Change point problems < Canonical change point problems < Fon<parametric and Bayesian approaches to the change<point problems < 2eedbac! in change point problems, .pproximations in bases and compression < &inear and non<linear approximations < 4arhunen &oeve approximations < )ransform coding < Distortion rate of #uanti?ation Sparse signal processing < Sparse representation and recovery, pursuit algorithms, compressive sensing, onte Carlo methods in signal processing < Particle filter *nference in +idden ar!ov odels, (xpectation< aximi?ation $( % algorithm' Text/References

Petre Stoica and "andolph

oses, --Spectral.nalysis of Signals--, Prentice +all, 6HHB'

Jacob Benesty, Jingdong Chen, and Iiteng+uang, -- icrophone array signal processing--,Springer, 6HH9' +arry &' 1an )ress, --;ptimum .rray Processing--, Part *1 of Detection,(stimation, and odulation )heory, ,iley *nterscience, 6HH6' Stefane 6H7H' allat, --. wavelet tour of signal processing5 the sparse way--, .cademic Press,

James Candy, --Bayesian Signal Processing--,,iley E Sons', 6HH8' ;livier Cappe, (ric oulines, and )obias "yden,--*nference in +idden Springer Series in Statistics, 6HHB' ar!ov odels--,

.rnaud Doucet and .' ' Johansen, --. )utorial on Particle 2iltering and Smoothing5 2ifteen years &ater--, in +andboo! of Fonlinear 2iltering$eds' D' Crisan et B' "o?ovs!y%, ;xford 0niversityPress, 6H77' I' C' (ldar and 3' 4utynio!, --Compressed Sensing5 )heory and .pplications --, Cambridge 0niversity Press, 6H77' "esearch publications that will be suggested during the course'

haptics wavelets < short<time 2) !alman theory linear algebra restoration segmentation enhancement mar!ov chains matrix computation

Two-dimensional Spectral Estimation 2. Image Smoothing with Roof and Step Edge Preserving, April 200 . !. Image Segmentation "ased on Tree- Str#ct#red $ar%ov Random &ields, April 2000. '. Estimation of Image (oise )ariance, April 2000.

*. +,namic Image +ata -ompression in Spatial and Temporal +omain, April .... /. $orphological Image Enhancement, April .... .. A#tomatic Segmentation of "rain $RI, April .... 0. Image Segmentation 0, E$ Algorithm Initialised 0, Tree Str#ct#re Scheme, April .... . $ar%ov Random &ield Segmentation of "rain $RI, April .... '. 1ossless Image compression #sing +-T and resid#als, April ../. 2. Image segmentation #sing f#33, c-means cl#stering algorithm, April ..*. /. Image data compression #sing d,adic transform, April ..2. .. Edge +etection in an Image 4sing the Rec#rsive 5avelet Transform, April ..'. 20. Acc#rate Image Rotation, April ..6.